[ReserveGrowthRulz]

I'm confused. Implied in your entire post is the idea that depletion alone determines production rate in the future. I have no problem whatsoever with the fundamental idea that once a well/field stabilizes at its initial production, there isn't much of anything left to do but go down.

This being the case, and this having been going on for the last 15 decades or so, why do you feel that suddenly it is the main driver for future production rates?

Depletion alone is only a part of the overall dynamics which go into worldwide production rate, without accounting for new discoveries, their sizes and future rates, old fields changing their depletion profile through better technology and reserve growth, new areas opening up to exploration as, say, the Arctic seapack melts and makes more areas available, without including all of these it would seem to me depletion modeling exclusively is kinda like closing your eyes, grabbing an elephant by the tail and trying to describe his size off of what you've got in your hand?

[ReserveGrowthRulz]

Well, I appreciate the depth of your answer, except for the little part where you didn't really seem to answer my question. By converting what I view as a basic question into a "gee the optimists are back" mission statement combined with a "I know statistics and public policy" broadside I figure maybe that means you don't have an answer?

This would be a surprise, because people who do work on the public policy side and know statistics and whatever other special and interesting knowledge you claim to have would have considered more than just the depletion rate of fields and wells and then claimed that the modelling of it alone explains away everything.

As to whether or not the world is going to end in...what did you say...the next 9 years or so(?)....well, that strikes me as an entirely different debate.

[ReserveGrowthRulz]

For starters, I have no problems with analyzing trends. Good trends, bad trends, whatever. I have no trouble quantifying trends using statistics, if only because its a common language to explain differences in data. And I don't feel that fitting whatever functions you have on hand to for depletion modelling is BAD...particularly since I use decline curve analysis all the time on a well by well basis.

My question stems from why someone would concentrate on only a small part of the problem, how fast things deplete, while casting that knowledge into the larger question, how soon do we all die off, and acting like it alone, or primarily, is the answer for when bad things happen during the PO phase of oil production?

As far as how smart you are with a degree in mathematics compared to just a run of the mill petroleum engineer, hey, I give you best. As far as I'm concerned high end mathmatics, while occasionally interesting, isn't near as much fun as turning more common mathematics into practical solutions, studies and knowledge which can be used by everyone today, tomorrow or the next day to make their lives a little better.

As far as your bet, I certainly won't take you up on it, but only because I'm not much of a gambling man, and I haven't satisfied myself yet what the new world order is going to look like with a minimum of excess capacity, something I don't think anyone has a good handle on quite yet, and something which I believe tends to alter the underlying dynamics of the entire game, particularly from an economics standpoint.

[rockdoc123]

Perhaps I can toss in my two pence worth as someone who is intimately familiar with oil and gas production and exploration and the proper use of stochaistic modeling as a predictive tool.

The underlying problem in taking datasets from various fields and using it to predict ultimate recovery is that not all data is created equally. Hubberts whole premise on this started out from the observation of oilfield engineers that production from a single well producing from a single zone would generally followed a decline trend..either geometric or hyperbolic. And indeed when you are working with fields producing from single zones this is an effective way of figuring out URR....although not as reliable say as material balance calculations (using pressure decline). The problem becomes if the production history in a pool or field is not that simple.

As an example in a given country there may be changes to fiscal regimes that determines fields previously economic are no longer economic...or as is the case in Algeria the gov't may change it's view on the rate that certain fields can be produced at. As a consequence you can see all sorts of wierd production curves that are double or triple peaked. As well as drilling continues we often see that new reservoirs become productive. An excellent example of this I think would be the Belayim Marine field in Egypts Gulf of Suez. The field produced for years from the Miocene Kareem formation ...reserves being in the order of 700 mmb to 1 gb. Quite a way into the history of the field a development well accidently struck hydrocarbons from a deeper reservoir...as it turns out this reservoir added another 500 MMB to the overall field. When you expand this sort of analysis up to the country level it encompasses the argument that Lynch has made regarding the North Sea or the Michigan basin. In both cases if you applied the Hubbert gausian analysis early in the history you would have missed the second hump of production....in UK due to changing tax regime and better economics for exploration and in the Michigan basin due to discovery of new play fairways. The method works well when you know how much more oil is left to be found...unfortunately that is somewhat of a moving target and indeed has some dependance on economics.
So as I have said before...the math makes sense but you cannot blindly treat subsurface information without some understanding of what you are dealing with. Often you are mixing apples with oranges which is never a good thing. The baysien approach is of course the best ....but my guess is it will simply show a p20-80 range that encompasses everything from a peak in 2010 to a peak in 2025....a pretty big range.

[ReserveGrowthRulz]

While I find all the talk of polynomials and goodness of fit and probability estimates all wildly exciting in an academic sort of way, I am the first to admit that, if people feel the need to use neato methods to model something, let'em do it, don't get in their way...and then don't forget to ask the basic question...does it matter, and was the effort worth another 0.0005% of accuracy...and does that accuracy MEAN anything in the greater scheme of answering the basic question.

While I might be fond of applied procedures to answer a question, its only because thats what I do. No one I work with cares much whether or not the answer is 2.0000 or 2.0001, for all practical purposes, 2 = 2 and will work as a solution set when the first thing you admit is that you don't know if 2.0000 is any better an approximation of the answer than 2.0001 is.

So before we get all caught up in shifted log normals and pareto's and whatever else can be dreamed up because we like it, lets start with the basics.

When you talk about "depletion", I get the feeling you are including alot more than what I would, as a petroleum engineer watching any one particular well "deplete". I never said and hope my post wasn't construed as me trying to equate individual well declines with predicting field declines. Could I do it? Sure...give me all the well production data you've got, and the time sequence, and I can assemble the entire mess into a field decline. It goes up, it usually peaks or plateau's, it declines. The fields profile ends up looking like all the individual wells' profile because thats how the fields profile is built.

Forecasting a field's decline WITHOUT knowing all the bits and parts and pieces that went into it adds a "shooting in the dark" component which we can all argue about until we wake up one morning and have a hydrogen economy.

When you talk about depleting something, are you referring to information gained at the individual level which you then scale up, or are you trying to estimate the QUALITY of your "shooting in the dark" estimate for depletion, admittedly a piece of the overall puzzle but hardly the only one?

[ReserveGrowthRulz]

Okay, so it appears that your first cut at data is to fit an equation to the data. Fine. Single polynomials, triples, sixth degree, whatever. Sounds not much different than decline curve analysis for reserves reporting. Except the accountants get cranky over any math they aren't familiar with.

However, I would think that just curve fitting doesn't actually predict or explain anything, its just best fitting a line to data. Without considering the bits and pieces that go into WHY things go up and down, isn't fitting trends just as good a shot in the dark for an answer as what Simmons does? Pick a position, show the data you like which supports your guess, and off to the races.

Using a trend projection means you are fitting THINGS which happened, some of which may happen again, some which may not, some of which were intentional ( oil embargo's ) and some of which weren't ( hurricanes knocking out production in the GOM, recessions ). So...once you've got a curve fitting routine which cycles through these events, it sure sounds like you have just "trendized" things which may or may not happen again, which may or may not happen at the same distance in time from each other that they did in the past, in general its just curve fitting, and while it fits the data to such and such a goodness of fact, or accounts for this much or that much of the data, it doesn't actually MEAN anything other than it fits the past okay. So...using the past to predict the future, how does your trend handle another Iranian revolution? Embargo? Worldwide recession? Electric cars in the States? Seems like any large event would negate any predictions you could make off the trend in the first place?

Don't take this as a putdown, it just seems like you are trying to use an equation to summarize the last 30 years of geopolitics, embargo's, wars, Saudia enforcement of quota's, and the equation has NOTHING in it related to those THINGS...it just hits the past data better than some other kind of equation, without explaining it, and without being able to count on it for predicting much of anything.

[ReserveGrowthRulz]

I'm a petroleum engineer. So yes, I've been through the calculus and the differential equation routine in college plus what I've needed along the way.

I don't argue that supply and demand don't work. I wouldn't argue with artificial things that can happen when a silly Cartel runs the table and US energy policy over that same time period has been lacking, if it exists at all.

I would argue that all the artificial influences over oil production are so large as to make mathematical modelling mostly meaningless. Its part of the reason why I figure the IEA gets it wrong nearly ALL the time...they literally have to wait until something happens before they react to it, and then all they are doing is reacting to past events, there isn't any more predictive value to their methods than any other math model which doesn't take into account the next unexpected event.

Another big booboo sitting smack in the middle of any modelling would be Saudi excess capacity, something they have had since the late-70s', and something which they have used to control price artificially. Not OPEC mind you, but the Saudi's. Using such excess capacity, they could pretty much do what they wished with the price of crude during the 80's and 90's, so modeling on anything related to price would have that particular anomaly sitting right in the middle of their data. So modeling price is more modeling the Saudi royal familys desire to open the taps more than anything else. Modeling demand is probably even worse, because demand can be influenced with price, and price has been manipulated up until pretty recently.

I think most Peakers get lost in the doom and gloom scenario's and get carried away with it, Campbell himself being the prime example. He must really love the attention, to be so wrong for so long and to STILL keep standing up and proclaiming the sky is falling....again...with a straight face.

[ReserveGrowthRulz]

It is unfortunate that you view my philosophical bent on the subject of Peak Oil as a slap against math in general, because it isn't.

Peakers use math to try and make their point ( the ones who aren't just arm waving Doomists anyway) , they do tend to get irritated when logic blows huge holes in assumptions fundamental to their case. The very idea that trend analysis can map a trend to the nth degree of certainty and have NOTHING to do with the underlying data and have ZERO ability beyond luck to be within a standard deviation of the answer 2 years from today is just the way things go sometimes when you assume the answer is contained in a 6th degree polynomial and not some crackpot zealot running Iran who decides to start a war one afternoon for fun.

As a petroleum engineer I am intimate with the idea that things produce, things deplete, and given a limited geologic resource which is produced to match an exponentially increasing demand, that something similar to Hubberts curve is as good a way to guess an future answer as any. Peak now, peak later, odds are, it will peak sometime. The problem is this....examples Hubbert used in his original paper went to crap in a handbasket when extraordinary economic pressures were applied, the specific example I'm thinking of is his production graph of Ohio. Hubbert also missed LOTS of neat and interesting things which have happened since the 50's and 60's, and I'm speculating here, but maybe thats why his world oil peak was wrong, maybe thats why his natural gas peak both national and world was wrong, because for a single, small, well delineated example, he could get CLOSE using his particular model, but the instant you try and extrapolate that model into the rest of the world? Which is different? More chaotic? Run mostly by unpredictable crackpots? The model falls apart, even if Campbell and Co have made a living proclaiming the sky is falling off a single, small, almost right prediction, while ignoring the problem of why only ONE was close, and the others were wrong. Do they take the lesson from that of "you ought to be careful extrapolating into the future when you only have a historical success ratio of 25%?" nope....they figure maybe others won't notice if they just rachet up the volume.

As to what kinds of curves I use? I'm usually pretty happy with hyperbolic and its related subsets, I haven't seen production data yet, well or fields, which requires anything much more complicated than that dictated by the geology and standard production practices of industry.

[WebHubbleTelescope]

Depletion alone is only a part of the overall dynamics which go into worldwide production rate, without accounting for new discoveries, their sizes and future rates, old fields changing their depletion profile through better technology and reserve growth, new areas opening up to exploration as, say, the Arctic seapack melts and makes more areas available, without including all of these it would seem to me depletion modeling exclusively is kinda like closing your eyes, grabbing an elephant by the tail and trying to describe his size off of what you've got in your hand?

Using the melting Arctic seapack as a rationalization for anything strikes me as a last gasp attempt at maintaining the status quo.

Cripes, if the seapack starts to melt at rates at which we need to replenish our oil supply we have a whole "boatload" of problems to start worrying about.

[rockdoc123]

Cripes, if the seapack starts to melt at rates at which we need to replenish our oil supply we have a whole "boatload" of problems to start worrying about.

To be fair I do not think that was his point at all (no mention was made about anything "saving" us from peakoil....seems to me all he is saying is that there are a number of underlying factors that affect production history in fields, basins, regions, countries and hence the world. If your model is based on past production history it only takes into account the particular factors that interacted during that time period....if suddenly other factors rear their head or become more influential then your model will likely have problems in its ability to predict a mean or even a p20-p80 range of outcomes. If you can come up with a model that predicts all of the potential interactions and resultant range of affects then great....my guess is we are looking at chaos theory at that point in time.